what's the problem with RIP

[biaobro]

Hi Team:

The top half is running OSPF, and the down half is running RIPv2 +  EIGRP.

R1 advertise 11.1.1.0 with RIPv2, R3 advertise 33.3.3.0 with EIGRP.

Then I redistribute both RIPv2 and EIGRP into OSPF in R2 and R4. no other setting on redistribute, which means all the routes will be O E2 type and Metric 20 int end router.

Finally I get route table from R5, something makes me confused that the routes to 33.3.3.0 formed ECMP but not for 11.1.1.0.  Routes to 11.1.1.0 always keep one either 25.1.1.2 or 45.1.1.4, the route type and metric are all same, but why no ECMP in 11.1.1.0? 

I attach the topo and config here, could someone help to explain? 

R1(config-router)#do show run | s router
router eigrp 100
network 12.1.1.1 0.0.0.0
network 13.1.1.1 0.0.0.0
router rip
version 2
network 11.0.0.0
network 12.0.0.0
network 13.0.0.0
no auto-summary

R2(config-router)#do show run | s router
router eigrp 100
network 12.1.1.2 0.0.0.0
router ospf 100
redistribute rip subnets
redistribute eigrp 100 subnets
network 25.1.1.2 0.0.0.0 area 0
router rip
version 2
network 12.0.0.0
no auto-summary

R3(config-router)#do show run | s router
router eigrp 100
network 13.1.1.3 0.0.0.0
network 33.3.3.3 0.0.0.0
network 34.1.1.3 0.0.0.0
router rip
version 2
network 13.0.0.0
network 34.0.0.0
no auto-summary

R4(config-router)#do show run | s router
router eigrp 100
network 34.1.1.4 0.0.0.0
router ospf 100
redistribute rip subnets
redistribute eigrp 100 subnets
network 45.1.1.4 0.0.0.0 area 0
router rip
version 2
network 34.0.0.0
no auto-summary

toporoute table

[Georg Pauwen]

Hello,

did you manipulate the interface cost ? As far as I can tell, the 'problem' is that there actually IS equal cost load balancing to 33.3.3.0, as there is an additional router in the path that should make the cost higher. Can you post the output of:

sh ip ospf interface Gi0/0 | in Cost

sh ip ospf interface Gi0/1 | in Cost

from both the R2 and R4 routers ?

[biaobro]

Hi Georg:

Thanks for the reply. I didn't manipulate interface cost. Except necessary protocol config and redistribution, all other settings are default where I think they also should be same in any interface. 

I checked the cost output as here :

[paul driver]

Hello
Prior to redistribution both ASBR R2/R4 will prefer the eigrp routes over rip so it’s basically down how they are seeing the loopback interfaces of R1/R3 Plus by default post redistribution as no specific metric is being defined ospf will append a default external cost of 20 to any external route also by default these external routes are costed as type E2 as such the internal cost to the ASBR (R2-R4) isn’t calculated, And looking at you configuration the difference is the way you are advertising the R1/R3 loopbacks:

R1 loopback0  is being advertised via RIP then redistributed 
R3 loopback0  is being advertised via eigrp then redistributed

[MHM Cisco World]

I under deep study for using same link for two or more routing process, until now I dont get nothing. 
but anyway 
if you do

show ip rip neighbor

are you see neighbor ?
I think NO
why because you could not use same link for two routing protocol (for now and if i get any infrom i will share it here).
so that explain the behave of R5. 

[paul driver]

Hello
why because you could not use same link for two routing protocol

I believe you can - by default the better of the two process pertaining to their administrative distance will take preference to which populates the rib table.

[MHM Cisco World]

I am not talk about advertise same prefix I talk use same net to establish two different routing protocol between same two router, here you can not. 

[paul driver]

I talk use same net to establish two different routing protocol between same two router, here you can not. 

maybe i’m misunderstanding your wording so apologies but you CAN establish or run two different routing protocols over the same network- what makes you think you cannot?

[biaobro]

Hi MHM:

show ip rip neighbor

  —— no neighbor list in output, but I think that's natural cause there is no neighbor in RIP, I'm not sure why CIsco provide this command. In lower version IOS, I only get

show ip rip database

I also built a very simple star topo to check the output , there is no output.

[paul driver]

Hello 

as a test redistribute the same routing process into ospf as metric-type 1

[MHM Cisco World]

I return to home run small lab
remember I dont like idea of run two protocol in one link 
so I delete the EIGRP 
Now return to your issue 
I see the same issue 
R5 only add one path to prefix advertise, 
I make some analysis and I found the issue 
the R5 select the path through R4 (45.0.0.4) and advertise to R2 
here the issue 
OSPF have AD 110 
RIP have 120 
so the R2 prefer the path through ospf (via R5) not through direct connect R1. 
solution 
change the AD of OSPF be more than 120. 

[paul driver]

hello @MHM Cisco World 
As per OP there is no path cost manipulation so from R5 perspective as it stands the redistribution is default E2 and the internal ospf calculation to either ASBR isn’t included also the redistributed rip/eigrp processes into ospf are default so the default costing will be 20.

as stated rtr1 loopback is via rip and rtr 3 is via eigrp - both asbr are eigrp aware so eigrp WILL take preference prior to redistribution into ospf hence i believe why the OP sees ecmp for r3 loopback 0 and not rtr 1 loopback 0

[MHM Cisco World · Accepted solution]

eigrp we see equal because AD is 90 and OSPF is 110 so 90 is lower 
for rip the ad is 120 and ospf is 110 and hence the issue appear. 

[biaobro]

That's true. I also noticed that. 

In database of R5, there is only one route to RIP destination.  

Is it just like that R5 told R2, I have better OSPF path(lower AD) to RIP destination, so R2 compare the path with local RIP path, then select OSPF path and not advertise RIP path to R5 anymore. 

[paul driver]

correct - as stated previous  ASBR R2/R4 areboth eigrp aware so will prefer the eigrp routes over rip prior to redistribution, into ospf via E2 lsa.( equal cost of 20)